LP capacity expansion problem benchmark

[kdheepak]

I have a .lp file generated from Pyomo (using openTEPES - https://github.com/IIT-EnergySystemModels/openTEPES/) that is an order of magnitude slower than a commercial solver, and was wondering if it'd be useful to include as a benchmark.

I've attached the file here (20mb zipped, 425mb unzipped): run1-lp-file.zip

It takes roughly 60 minutes to run on a Mac with HiGHS. On Windows, this hangs indefinitely. I left it running for a few hours and then killed the process, I believe the hanging is related to this issue: ERGO-Code/HiGHS#1044.

With a commercial solver, this lp file solves in a few minutes.

fwiw there's a smaller problem with the same formulation (i.e. same constraints but over fewer time periods - 1 week instead of 1 year) that you can run in a few minutes with HiGHS here: https://github.com/IIT-EnergySystemModels/openTEPES/blob/6f70339e11daaea3f372c06e035dd9a021b6d4b1/tests/test_run.py

[joaquimg]

Thanks! Our focus will be mps files that we can generate with JuMP, but this will be useful.

[erling-d-andersen]

Interesting. However, it seems the run1.lp files has no integer variables.

[odow]

I renamed the title of this issue